摘要
针对目前甘蔗机械化播种难以满足蔗种蔗芽水平朝向种沟沟壁的种植农艺要求,本文研究蔗芽识别与蔗种位姿调整算法,并设计基于机器视觉的蔗芽识别与蔗种位姿调整系统。基于灰度变换算法对蔗种图像进行增强处理,采用Otsu算法获取蔗芽阈值参数,分割出蔗芽完整轮廓;基于二值图像模式下的几何特征提取蔗芽位置特征参数,利用最小二乘线性拟合法提取蔗种边缘轮廓;根据蔗芽位置与形状特征参数,构建蔗种空间位姿模型。计算蔗芽位姿,由机械手夹持蔗种将其调整到理想位姿。选取80幅蔗种图像作为试验样本,试验结果表明蔗芽位姿识别准确率93.75%,定向成功率94.5%。调整后蔗芽位姿基本贴近水平面,蔗芽水平朝向种沟两侧,满足蔗种自动定向播种农艺要求。本研究可为蔗种机械化定向播种提供解决方案,并为甘蔗种植机的自动定向装置设计提供参考。
Aiming at the problem that the current automatic sugarcane seeding is difficult to meet the agronomic requirements of"sugarcane buds face both sides of the seed furrow level",a feasible system was designed to realize the recognition and posture adjustment of sugarcane buds using machine vision,and the algorithms are studied to recognize and adjust posture of sugarcane buds.The sugarcane seed images are enhanced based on the gray-scale transformation algorithm,the Otsu algorithm is used to obtain the sugarcane buds threshold parameters and segment the sugarcane buds with complete contours.The sugarcane buds position feature parameters are extracted based on the geometric characteristics in the binary image mode,and the shape feature parameters are extracted using least squares linear fitting method;the posture of the sugarcane buds are calculated according to the position and shape feature parameters of the sugarcane buds,and the sugarcane seeds are adjusted to the ideal posture by the directional adjusting device.80 images were selected as the test samples,the test results showed that the accuracy of sugarcane buds pose recognition was 93.75%,the recognition speed was 0.6 s/frame,and the success rate of orientation adjustment was 94.5%.After adjustment,the position of the sugarcane buds is basically close to the horizontal plane,which meets the agronomic requirements of automatic orientation planting of sugarcane seeds.This research provides a solution for the automatic orientation of sugarcane seeds and also provides a reference for the design of the automatic orientation device of the sugarcane planting machine.
作者
许洪振
王明明
陈海锦
刘姣娣
XU Hongzhen;WANG Mingming;CHEN Haijin;LIU Jiaodi(College of Mechanical and Control Engineering,Guilin University of Technology,Guilin,Guangxi 541006,China)
出处
《石河子大学学报（自然科学版）》
CAS
北大核心
2023年第2期154-159,共6页
Journal of Shihezi University(Natural Science)
基金
国家自然科学基金(52265028)
广西省自然科学基金(2021JJA160046)
广西研究生教育创新计划(YCSW2022333)
关键词
蔗芽
机器视觉
位姿识别
定向
sugarcane buds
computer vision
posture recognition
orientation
